Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add Azzygoatcoder/agent-useful-skills --skill self-evolvegit clone --depth 1 https://github.com/Azzygoatcoder/agent-useful-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/azzygoatcoder/agent-useful-skills/self-evolve)<a href="https://agentmods.dev/skills/azzygoatcoder/agent-useful-skills/self-evolve"><img src="https://agentmods.dev/badge/skills/azzygoatcoder/agent-useful-skills/self-evolve.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00080 | $0.01198 |
| Opus 5 | $0.00040 | $0.00599 |
| Sonnet 5 | $0.00016 | $0.00240 |
| Haiku 4.5 | $0.00008 | $0.00120 |
Grade A, and why
self-evolve scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Self-Evolve — 自进化闭环
把「自进化日志」从被动记录变成主动闭环：定期读日志 → 判断哪些教训该晋升为核心流程 → 落地改 SKILL.md → 记「已晋升」。
场景判定
| 场景 | 触发 | 动作 |
|---|---|---|
| A. 汇总 | "consolidate"/"看看所有日志" | grep 所有 skill 日志 → 汇总视图 |
| B. 单 skill 进化 | "进化 paper-reading"/"X 的日志该晋升了吗" | 读该 skill 核心流程 + 日志 → 提晋升建议 |
| C. 顺带检查 | 改完一个 skill 后 | 记日志一行；同类教训重复 ≥2 次且有独立验证 → 提示晋升 |
场景 B：单 skill 进化（核心）
- 读：该 skill 的
## 自进化日志表 + 核心流程（场景判定表 / 工作流步骤） - 判：逐条日志过两道门槛——① 独立验证（跨模型/真机/人）；② 事实/补偿分类（🟢环境事实 / 🟡模型补偿 / 🔴偶然）——
- 晋升：🟢 且有独立验证，但核心流程还没写 → 写进 SKILL.md 主体
- 降级：🟡 模型补偿 → 只留该模型作用域，不进通用 SKILL.md
- 过时：核心流程已更新、此条已被吸收 → 从日志删或标记
- 保留：🔴 一次性 / 待验证 → 留日志
- 提：产出「晋升/过时」的具体改法（贴 SKILL.md 片段 diff，不整篇重写）
- 改：人工确认后走 writing-skills 流程落地 → 验证
- 记：日志加一行
已晋升 | 日期 | 把 X 写进核心流程
晋升判据（反自举：独立验证 + 事实/补偿分类）
两条门槛都要过：
- 独立验证：教训被异源验证过——跨模型复核（review.py / vision）通过，或真机/真命令复现（node --check、真实场景试跑），或人类确认。"出现 ≥2 次"不算独立验证——可能是同一个盲点犯了两次。
- 事实/补偿分类：先打标签再决定去向——
- 🟢 环境事实 → 晋升进通用 SKILL.md：与模型无关的客观规律（平台行为、文件系统语义、命令行为，如"Windows junction 在 Node readdir 里 isDirectory()=false"）。
- 🟡 模型补偿 → 不晋升通用：特定模型的权宜（如"DeepSeek 跟随差所以要列 20 个触发词"）。只留在该模型作用域（描述/profile），绝不写进通用 SKILL.md——否则就是把模型训练成只认自家格式的刻板化。
- 🔴 一次性偶然 → 不晋升。
- 不晋升：无独立验证、模型行为补偿（除非限定该模型作用域）、单次偶然。
- 衰减：已晋升规则长期无真实使用/引用（接 Skill-BOM：宣称 vs 实际）→ 标记候选退役，防规则无限膨胀。
工具
| 工具 | 用途 |
|---|---|
Grep ## 自进化日志 |
定位所有 skill 的日志 |
| writing-skills skill | 落地改 SKILL.md |
| review.py（可选） | 跨模型复核——晋升判据的「独立验证」门槛之一 |
自进化日志
| 日期 | 学习来源 | 吸收的模式 |
|---|---|---|
| 2026-08-13 | 首版 | 自进化闭环 v1：汇总/进化/顺带三场景；晋升判据=复用 ≥2 次或核心流程空白 |
| 2026-08-14 | 自举反思（Claude Code bootstrap 教训） | 晋升判据 v2：独立验证门槛（跨模型/真机/人，≥2 次不算）+ 事实/补偿分类（🟢环境事实晋升 / 🟡模型补偿只留模型作用域 / 🔴偶然）+ 衰减机制（接 Skill-BOM）——防自消耗坍缩与工具方言刻板 |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 7d ago First seen · 57 lines · 80 tokens per session scan A b208ab2a42cf
self-evolve is a skill published in the GitHub repository Azzygoatcoder/agent-useful-skills (6 stars, last pushed 9d ago), licensed MIT. It adds 80 tokens to every session and 1,198 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
figure-rhetoric
Evaluate whether figures and plots in a manuscript effectively communicate the claims they support. Audits chart-type fit, axis design, visual hierarchy, data density, caption interpretation, perceptual accuracy, and narrative arc across 8 dimensions. Triggers on: "do my figures work", "check my plots", "are my graphs…
paper-planning
Guides pre-writing planning for academic papers with 4 structured steps: story design (task-challenge-insight-contribution-advantage), experiment planning (comparisons + ablations), figure design (pipeline + teaser), and 4-week timeline management. Includes counterintuitive planning tactics (write a mock rejection…
arxiv-package
Package a TeX/LaTeX project into a clean tarball or zip for arXiv upload: file selection, build-artifact exclusion, 00README.XXX generation, ancillary file organization, archive validation. Triggers on: "package for arXiv", "create arXiv tarball", "bundle submission", "zip for arXiv", "prepare arXiv upload", "arXiv…
paper-review
Guides self-review of YOUR OWN academic paper before submission with adversarial stress-testing. Core method: 5-aspect checklist (contribution sufficiency, writing clarity, results quality, testing completeness, method design), counterintuitive protocol (reject-first simulation, delete unsupported claims, score trust…
research-critique
Critical analysis of research papers evaluating methodology, claims-evidence alignment, and contribution significance. Triggers on: "critique this paper", "review this research", "analyze this study", "evaluate the methodology", "is this paper sound". NOT for formatting or submission readiness, use manuscript-review.
paper2poster
Convert academic papers (PDF) into conference posters (HTML/PNG). You are the conductor: you decide what each section needs — an original paper figure or text — write the outline, hand-author the poster HTML, and iterate on the render using your own visual read and a blind-reader content quiz. Use when the user wants…